文章目录
Mamba_4">引言：Mamba
2024年04月29日16:06:08，今天开始记录mamba模块的学习与使用过程。
第一章：环境安装
亲测，根据下文的安装步骤，
即可成功！
使用代码Vision Mamba：https://github.com/hustvl/Vim
git clone https://github.com/hustvl/Vim.git
1.1安装教程
安装教程：下载好vision mamba后，根据下面的教程一步一步安装即可成功。
vision mamba 运行训练记录，解决bimamba_type错误
1.2问题总结
问题总结：遇见的问题可以参考这个链接，总结的比较全面。
1.3安装总结
关键就是下载causal_conv1d和mamba_ssm，最好是下载离线的whl文件，然后再用pip进行安装。值得注意的一点就是要用官方项目里的mamba_ssm替换安装在conda环境里的mamba_ssm。
第二章：即插即用模块
Mamba_Vision_38">2.1模块一：Mamba Vision
Github：https://github.com/hustvl/Vim；
下载代码，配置好环境后，用下面的代码替换Vim/vim/models_mamba.py，即可直接运行；
运行指令
python models_mamba.py
代码：models_mamba.py
# Copyright (c) 2015-present, Facebook, Inc.
# All rights reserved.
import torch
import torch.nn as nn
from functools import partial
from torch import Tensor
from typing import Optionalfrom timm.models.vision_transformer import VisionTransformer, _cfg
from timm.models.registry import register_model
from timm.models.layers import trunc_normal_, lecun_normal_from timm.models.layers import DropPath, to_2tuple
from timm.models.vision_transformer import _load_weightsimport mathfrom collections import namedtuplefrom mamba_ssm.modules.mamba_simple import Mamba
from mamba_ssm.utils.generation import GenerationMixin
from mamba_ssm.utils.hf import load_config_hf, load_state_dict_hffrom rope import *
import randomtry:from mamba_ssm.ops.triton.layernorm import RMSNorm, layer_norm_fn, rms_norm_fn
except ImportError:RMSNorm, layer_norm_fn, rms_norm_fn = None, None, None__all__ = ['vim_tiny_patch16_224', 'vim_small_patch16_224', 'vim_base_patch16_224','vim_tiny_patch16_384', 'vim_small_patch16_384', 'vim_base_patch16_384',
]class PatchEmbed(nn.Module):""" 2D Image to Patch Embedding"""def __init__(self, img_size=224, patch_size=16, stride=16, in_chans=3, embed_dim=768, norm_layer=None, flatten=True):super().__init__()img_size = to_2tuple(img_size)patch_size = to_2tuple(patch_size)self.img_size = img_sizeself.patch_size = patch_sizeself.grid_size = ((img_size[0] - patch_size[0]) // stride + 1, (img_size[1] - patch_size[1]) // stride + 1)self.num_patches = self.grid_size[0] * self.grid_size[1]self.flatten = flattenself.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=stride)self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()def forward(self, x):B, C, H, W = x.shapeassert H == self.img_size[0] and W == self.img_size[1], \f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})."x = self.proj(x)if self.flatten:x = x.flatten(2).transpose(1, 2) # BCHW -> BNCx = self.norm(x)return xclass Block(nn.Module):def __init__(self, dim, mixer_cls, norm_cls=nn.LayerNorm, fused_add_norm=False, residual_in_fp32=False,drop_path=0.,):"""Simple block wrapping a mixer class with LayerNorm/RMSNorm and residual connection"This Block has a slightly different structure compared to a regularprenorm Transformer block.The standard block is: LN -> MHA/MLP -> Add.[Ref: https://arxiv.org/abs/2002.04745]Here we have: Add -> LN -> Mixer, returning boththe hidden_states (output of the mixer) and the residual.This is purely for performance reasons, as we can fuse add and LayerNorm.The residual needs to be provided (except for the very first block)."""super().__init__()self.residual_in_fp32 = residual_in_fp32self.fused_add_norm = fused_add_normself.mixer = mixer_cls(dim)self.norm = norm_cls(dim)self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()if self.fused_add_norm:assert RMSNorm is not None, "RMSNorm import fails"assert isinstance(self.norm, (nn.LayerNorm, RMSNorm)), "Only LayerNorm and RMSNorm are supported for fused_add_norm"def forward(self, hidden_states: Tensor, residual: Optional[Tensor] = None, inference_params=None):r"""Pass the input through the encoder layer.Args:hidden_states: the sequence to the encoder layer (required).residual: hidden_states = Mixer(LN(residual))"""if not self.fused_add_norm:if residual is None:residual = hidden_stateselse:residual = residual + self.drop_path(hidden_states)hidden_states = self.norm(residual.to(dtype=self.norm.weight.dtype))if self.residual_in_fp32:residual = residual.to(torch.float32)else:fused_add_norm_fn = rms_norm_fn if isinstance(self.norm, RMSNorm) else layer_norm_fnif residual is None:hidden_states, residual = fused_add_norm_fn(hidden_states,self.norm.weight,self.norm.bias,residual=residual,prenorm=True,residual_in_fp32=self.residual_in_fp32,eps=self.norm.eps,)else:hidden_states, residual = fused_add_norm_fn(self.drop_path(hidden_states),self.norm.weight,self.norm.bias,residual=residual,prenorm=True,residual_in_fp32=self.residual_in_fp32,eps=self.norm.eps,) hidden_states = self.mixer(hidden_states, inference_params=inference_params)return hidden_states, residualdef allocate_inference_cache(self, batch_size, max_seqlen, dtype=None, **kwargs):return self.mixer.allocate_inference_cache(batch_size, max_seqlen, dtype=dtype, **kwargs)def create_block(d_model,ssm_cfg=None,norm_epsilon=1e-5,drop_path=0.,rms_norm=False,residual_in_fp32=False,fused_add_norm=False,layer_idx=None,device=None,dtype=None,if_bimamba=False,bimamba_type="none",if_devide_out=False,init_layer_scale=None,
):if if_bimamba:bimamba_type = "v1"if ssm_cfg is None:ssm_cfg = {}factory_kwargs = {"device": device, "dtype": dtype}mixer_cls = partial(Mamba, layer_idx=layer_idx, bimamba_type=bimamba_type, if_devide_out=if_devide_out, init_layer_scale=init_layer_scale, **ssm_cfg, **factory_kwargs)norm_cls = partial(nn.LayerNorm if not rms_norm else RMSNorm, eps=norm_epsilon, **factory_kwargs)block = Block(d_model,mixer_cls,norm_cls=norm_cls,drop_path=drop_path,fused_add_norm=fused_add_norm,residual_in_fp32=residual_in_fp32,)block.layer_idx = layer_idxreturn block# https://github.com/huggingface/transformers/blob/c28d04e9e252a1a099944e325685f14d242ecdcd/src/transformers/models/gpt2/modeling_gpt2.py#L454
def _init_weights(module,n_layer,initializer_range=0.02, # Now only used for embedding layer.rescale_prenorm_residual=True,n_residuals_per_layer=1, # Change to 2 if we have MLP
):if isinstance(module, nn.Linear):if module.bias is not None:if not getattr(module.bias, "_no_reinit", False):nn.init.zeros_(module.bias)elif isinstance(module, nn.Embedding):nn.init.normal_(module.weight, std=initializer_range)if rescale_prenorm_residual:# Reinitialize selected weights subject to the OpenAI GPT-2 Paper Scheme:# > A modified initialization which accounts for the accumulation on the residual path with model depth. Scale# > the weights of residual layers at initialization by a factor of 1/√N where N is the # of residual layers.# > -- GPT-2 :: https://openai.com/blog/better-language-models/## Reference (Megatron-LM): https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/model/gpt_model.pyfor name, p in module.named_parameters():if name in ["out_proj.weight", "fc2.weight"]:# Special Scaled Initialization --> There are 2 Layer Norms per Transformer Block# Following Pytorch init, except scale by 1/sqrt(2 * n_layer)# We need to reinit p since this code could be called multiple times# Having just p *= scale would repeatedly scale it downnn.init.kaiming_uniform_(p, a=math.sqrt(5))with torch.no_grad():p /= math.sqrt(n_residuals_per_layer * n_layer)def segm_init_weights(m):if isinstance(m, nn.Linear):trunc_normal_(m.weight, std=0.02)if isinstance(m, nn.Linear) and m.bias is not None:nn.init.constant_(m.bias, 0)elif isinstance(m, nn.Conv2d):# NOTE conv was left to pytorch default in my original initlecun_normal_(m.weight)if m.bias is not None:nn.init.zeros_(m.bias)elif isinstance(m, (nn.LayerNorm, nn.GroupNorm, nn.BatchNorm2d)):nn.init.zeros_(m.bias)nn.init.ones_(m.weight)class VisionMamba(nn.Module):def __init__(self, img_size=224, patch_size=16, stride=16,depth=24, embed_dim=192, channels=3, num_classes=1000,ssm_cfg=None, drop_rate=0.,drop_path_rate=0.1,norm_epsilon: float = 1e-5, rms_norm: bool = False, initializer_cfg=None,fused_add_norm=False,residual_in_fp32=False,device=None,dtype=None,ft_seq_len=None,pt_hw_seq_len=14,if_bidirectional=False,final_pool_type='none',if_abs_pos_embed=False,if_rope=False,if_rope_residual=False,flip_img_sequences_ratio=-1.,if_bimamba=False,bimamba_type="none",if_cls_token=False,if_devide_out=False,init_layer_scale=None,use_double_cls_token=False,use_middle_cls_token=False,**kwargs):factory_kwargs = {"device": device, "dtype": dtype}# add factory_kwargs into kwargskwargs.update(factory_kwargs) super().__init__()self.residual_in_fp32 = residual_in_fp32self.fused_add_norm = fused_add_normself.if_bidirectional = if_bidirectionalself.final_pool_type = final_pool_typeself.if_abs_pos_embed = if_abs_pos_embedself.if_rope = if_ropeself.if_rope_residual = if_rope_residualself.flip_img_sequences_ratio = flip_img_sequences_ratioself.if_cls_token = if_cls_tokenself.use_double_cls_token = use_double_cls_tokenself.use_middle_cls_token = use_middle_cls_tokenself.num_tokens = 1 if if_cls_token else 0# pretrain parametersself.num_classes = num_classesself.d_model = self.num_features = self.embed_dim = embed_dim # num_features for consistency with other modelsself.patch_embed = PatchEmbed(img_size=img_size, patch_size=patch_size, stride=stride, in_chans=channels, embed_dim=embed_dim)num_patches = self.patch_embed.num_patchesif if_cls_token:if use_double_cls_token:self.cls_token_head = nn.Parameter(torch.zeros(1, 1, self.embed_dim))self.cls_token_tail = nn.Parameter(torch.zeros(1, 1, self.embed_dim))self.num_tokens = 2else:self.cls_token = nn.Parameter(torch.zeros(1, 1, self.embed_dim))# self.num_tokens = 1if if_abs_pos_embed:self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + self.num_tokens, self.embed_dim))self.pos_drop = nn.Dropout(p=drop_rate)if if_rope:half_head_dim = embed_dim // 2hw_seq_len = img_size // patch_sizeself.rope = VisionRotaryEmbeddingFast(dim=half_head_dim,pt_seq_len=pt_hw_seq_len,ft_seq_len=hw_seq_len)self.head = nn.Linear(self.num_features, num_classes) if num_classes > 0 else nn.Identity()# TODO: release this commentdpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)] # stochastic depth decay rule# import ipdb;ipdb.set_trace()inter_dpr = [0.0] + dprself.drop_path = DropPath(drop_path_rate) if drop_path_rate > 0. else nn.Identity()# transformer blocksself.layers = nn.ModuleList([create_block(embed_dim,ssm_cfg=ssm_cfg,norm_epsilon=norm_epsilon,rms_norm=rms_norm,residual_in_fp32=residual_in_fp32,fused_add_norm=fused_add_norm,layer_idx=i,if_bimamba=if_bimamba,bimamba_type=bimamba_type,drop_path=inter_dpr[i],if_devide_out=if_devide_out,init_layer_scale=init_layer_scale,**factory_kwargs,)for i in range(depth)])# output headself.norm_f = (nn.LayerNorm if not rms_norm else RMSNorm)(embed_dim, eps=norm_epsilon, **factory_kwargs)# self.pre_logits = nn.Identity()# original initself.patch_embed.apply(segm_init_weights)self.head.apply(segm_init_weights)if if_abs_pos_embed:trunc_normal_(self.pos_embed, std=.02)if if_cls_token:if use_double_cls_token:trunc_normal_(self.cls_token_head, std=.02)trunc_normal_(self.cls_token_tail, std=.02)else:trunc_normal_(self.cls_token, std=.02)# mamba initself.apply(partial(_init_weights,n_layer=depth,**(initializer_cfg if initializer_cfg is not None else {}),))def allocate_inference_cache(self, batch_size, max_seqlen, dtype=None, **kwargs):return {i: layer.allocate_inference_cache(batch_size, max_seqlen, dtype=dtype, **kwargs)for i, layer in enumerate(self.layers)}@torch.jit.ignoredef no_weight_decay(self):return {"pos_embed", "cls_token", "dist_token", "cls_token_head", "cls_token_tail"}@torch.jit.ignore()def load_pretrained(self, checkpoint_path, prefix=""):_load_weights(self, checkpoint_path, prefix)def forward_features(self, x, inference_params=None, if_random_cls_token_position=False, if_random_token_rank=False):# taken from https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py# with slight modifications to add the dist_tokenx = self.patch_embed(x)B, M, _ = x.shapeif self.if_cls_token:if self.use_double_cls_token:cls_token_head = self.cls_token_head.expand(B, -1, -1)cls_token_tail = self.cls_token_tail.expand(B, -1, -1)token_position = [0, M + 1]x = torch.cat((cls_token_head, x, cls_token_tail), dim=1)M = x.shape[1]else:if self.use_middle_cls_token:cls_token = self.cls_token.expand(B, -1, -1)token_position = M // 2# add cls token in the middlex = torch.cat((x[:, :token_position, :], cls_token, x[:, token_position:, :]), dim=1)elif if_random_cls_token_position:cls_token = self.cls_token.expand(B, -1, -1)token_position = random.randint(0, M)x = torch.cat((x[:, :token_position, :], cls_token, x[:, token_position:, :]), dim=1)print("token_position: ", token_position)else:cls_token = self.cls_token.expand(B, -1, -1) # stole cls_tokens impl from Phil Wang, thankstoken_position = 0x = torch.cat((cls_token, x), dim=1)M = x.shape[1]if self.if_abs_pos_embed:# if new_grid_size[0] == self.patch_embed.grid_size[0] and new_grid_size[1] == self.patch_embed.grid_size[1]:# x = x + self.pos_embed# else:# pos_embed = interpolate_pos_embed_online(# self.pos_embed, self.patch_embed.grid_size, new_grid_size,0# )x = x + self.pos_embedx = self.pos_drop(x)if if_random_token_rank:# 生成随机 shuffle 索引shuffle_indices = torch.randperm(M)if isinstance(token_position, list):print("original value: ", x[0, token_position[0], 0], x[0, token_position[1], 0])else:print("original value: ", x[0, token_position, 0])print("original token_position: ", token_position)# 执行 shufflex = x[:, shuffle_indices, :]if isinstance(token_position, list):# 找到 cls token 在 shuffle 之后的新位置new_token_position = [torch.where(shuffle_indices == token_position[i])[0].item() for i in range(len(token_position))]token_position = new_token_positionelse:# 找到 cls token 在 shuffle 之后的新位置token_position = torch.where(shuffle_indices == token_position)[0].item()if isinstance(token_position, list):print("new value: ", x[0, token_position[0], 0], x[0, token_position[1], 0])else:print("new value: ", x[0, token_position, 0])print("new token_position: ", token_position)if_flip_img_sequences = Falseif self.flip_img_sequences_ratio > 0 and (self.flip_img_sequences_ratio - random.random()) > 1e-5:x = x.flip([1])if_flip_img_sequences = True# mamba implresidual = Nonehidden_states = xif not self.if_bidirectional:for layer in self.layers:if if_flip_img_sequences and self.if_rope:hidden_states = hidden_states.flip([1])if residual is not None:residual = residual.flip([1])# rope aboutif self.if_rope:hidden_states = self.rope(hidden_states)if residual is not None and self.if_rope_residual:residual = self.rope(residual)if if_flip_img_sequences and self.if_rope:hidden_states = hidden_states.flip([1])if residual is not None:residual = residual.flip([1])hidden_states, residual = layer(hidden_states, residual, inference_params=inference_params)else:# get two layers in a single for-loopfor i in range(len(self.layers) // 2):if self.if_rope:hidden_states = self.rope(hidden_states)if residual is not None and self.if_rope_residual:residual = self.rope(residual)hidden_states_f, residual_f = self.layers[i * 2](hidden_states, residual, inference_params=inference_params)hidden_states_b, residual_b = self.layers[i * 2 + 1](hidden_states.flip([1]), None if residual == None else residual.flip([1]), inference_params=inference_params)hidden_states = hidden_states_f + hidden_states_b.flip([1])residual = residual_f + residual_b.flip([1])if not self.fused_add_norm:if residual is None:residual = hidden_stateselse:residual = residual + self.drop_path(hidden_states)hidden_states = self.norm_f(residual.to(dtype=self.norm_f.weight.dtype))else:# Set prenorm=False here since we don't need the residualfused_add_norm_fn = rms_norm_fn if isinstance(self.norm_f, RMSNorm) else layer_norm_fnhidden_states = fused_add_norm_fn(self.drop_path(hidden_states),self.norm_f.weight,self.norm_f.bias,eps=self.norm_f.eps,residual=residual,prenorm=False,residual_in_fp32=self.residual_in_fp32,)# return only cls token if it existsif self.if_cls_token:if self.use_double_cls_token:return (hidden_states[:, token_position[0], :] + hidden_states[:, token_position[1], :]) / 2else:if self.use_middle_cls_token:return hidden_states[:, token_position, :]elif if_random_cls_token_position:return hidden_states[:, token_position, :]else:return hidden_states[:, token_position, :]if self.final_pool_type == 'none':return hidden_states[:, -1, :]elif self.final_pool_type == 'mean':return hidden_states.mean(dim=1)elif self.final_pool_type == 'max':return hidden_stateselif self.final_pool_type == 'all':return hidden_stateselse:raise NotImplementedErrordef forward(self, x, return_features=False, inference_params=None, if_random_cls_token_position=False, if_random_token_rank=False):x = self.forward_features(x, inference_params, if_random_cls_token_position=if_random_cls_token_position, if_random_token_rank=if_random_token_rank)if return_features:return xx = self.head(x)if self.final_pool_type == 'max':x = x.max(dim=1)[0]return x@register_model
def vim_tiny_patch16_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2(pretrained=False, **kwargs):model = VisionMamba(patch_size=16, embed_dim=192, depth=24, rms_norm=True, residual_in_fp32=True, fused_add_norm=True, final_pool_type='mean', if_abs_pos_embed=True, if_rope=False, if_rope_residual=False, bimamba_type="v2", if_cls_token=True, if_devide_out=True, use_middle_cls_token=True, **kwargs)model.default_cfg = _cfg()if pretrained:checkpoint = torch.hub.load_state_dict_from_url(url="to.do",map_location="cpu", check_hash=True)model.load_state_dict(checkpoint["model"])return model@register_model
def vim_tiny_patch16_stride8_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2(pretrained=False, **kwargs):model = VisionMamba(patch_size=16, stride=8, embed_dim=192, depth=24, rms_norm=True, residual_in_fp32=True, fused_add_norm=True, final_pool_type='mean', if_abs_pos_embed=True, if_rope=False, if_rope_residual=False, bimamba_type="v2", if_cls_token=True, if_devide_out=True, use_middle_cls_token=True, **kwargs)model.default_cfg = _cfg()if pretrained:checkpoint = torch.hub.load_state_dict_from_url(url="to.do",map_location="cpu", check_hash=True)model.load_state_dict(checkpoint["model"])return model@register_model
def vim_small_patch16_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2(pretrained=False, **kwargs):model = VisionMamba(patch_size=16, embed_dim=384, depth=24, rms_norm=True, residual_in_fp32=True, fused_add_norm=True, final_pool_type='mean', if_abs_pos_embed=True, if_rope=False, if_rope_residual=False, bimamba_type="v2", if_cls_token=True, if_devide_out=True, use_middle_cls_token=True, **kwargs)model.default_cfg = _cfg()if pretrained:checkpoint = torch.hub.load_state_dict_from_url(url="to.do",map_location="cpu", check_hash=True)model.load_state_dict(checkpoint["model"])return model@register_model
def vim_small_patch16_stride8_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2(pretrained=False, **kwargs):model = VisionMamba(patch_size=16, stride=8, embed_dim=384, depth=24, rms_norm=True, residual_in_fp32=True, fused_add_norm=True, final_pool_type='mean', if_abs_pos_embed=True, if_rope=False, if_rope_residual=False, bimamba_type="v2", if_cls_token=True, if_devide_out=True, use_middle_cls_token=True, **kwargs)model.default_cfg = _cfg()if pretrained:checkpoint = torch.hub.load_state_dict_from_url(url="to.do",map_location="cpu", check_hash=True)model.load_state_dict(checkpoint["model"])return modelif __name__ == '__main__':# cuda or cpudevice = torch.device("cuda" if torch.cuda.is_available() else "cpu")print(device)# 实例化模型得到分类结果inputs = torch.randn(1, 3, 224, 224).to(device)model = vim_small_patch16_stride8_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2(pretrained=False).to(device)print(model)outputs = model(inputs)print(outputs.shape)# 实例化mamba模块，输入输出特征维度不变 B C H Wx = torch.rand(10, 16, 64, 128).to(device)B, C, H, W = x.shapeprint("输入特征维度：", x.shape)x = x.view(B, C, H * W).permute(0, 2, 1)print("维度变换：", x.shape)mamba = create_block(d_model=C).to(device)# mamba模型代码中返回的是一个元组：hidden_states, residualhidden_states, residual = mamba(x)x = hidden_states.permute(0, 2, 1).view(B, C, H, W)print("输出特征维度：", x.shape)
运行结果
MambaIR_682">2.2模块二：MambaIR
B站UP主：@箫张跋扈
视频地址：Mamba Back！一种来自于Mamba领域的即插即用模块(TimeMachine)，用于时间序列任务！
下载好代码后，把下面的代码放到MambaIR.py文件中，然后再运行即可得到结果。
MambaIR_689">代码：MambaIR
# Code Implementation of the MambaIR Model
import warnings
warnings.filterwarnings("ignore")
import math
import torch
import torch.nn as nn
import torch.nn.functional as F
from functools import partial
from typing import Optional, Callable
from timm.models.layers import DropPath, to_2tuple, trunc_normal_
from mamba_ssm.ops.selective_scan_interface import selective_scan_fn, selective_scan_ref
from einops import rearrange, repeat"""
最近，选择性结构化状态空间模型，特别是改进版本的Mamba，在线性复杂度的远程依赖建模方面表现出了巨大的潜力。
然而，标准Mamba在低级视觉方面仍然面临一定的挑战，例如局部像素遗忘和通道冗余。在这项工作中，我们引入了局部增强和通道注意力来改进普通 Mamba。
通过这种方式，我们利用了局部像素相似性并减少了通道冗余。大量的实验证明了我们方法的优越性。
"""NEG_INF = -1000000class ChannelAttention(nn.Module):"""Channel attention used in RCAN.Args:num_feat (int): Channel number of intermediate features.squeeze_factor (int): Channel squeeze factor. Default: 16."""def __init__(self, num_feat, squeeze_factor=16):super(ChannelAttention, self).__init__()self.attention = nn.Sequential(nn.AdaptiveAvgPool2d(1),nn.Conv2d(num_feat, num_feat // squeeze_factor, 1, padding=0),nn.ReLU(inplace=True),nn.Conv2d(num_feat // squeeze_factor, num_feat, 1, padding=0),nn.Sigmoid())def forward(self, x):y = self.attention(x)return x * yclass CAB(nn.Module):def __init__(self, num_feat, is_light_sr= False, compress_ratio=3,squeeze_factor=30):super(CAB, self).__init__()if is_light_sr: # we use depth-wise conv for light-SR to achieve more efficientself.cab = nn.Sequential(nn.Conv2d(num_feat, num_feat, 3, 1, 1, groups=num_feat),ChannelAttention(num_feat, squeeze_factor))else: # for classic SRself.cab = nn.Sequential(nn.Conv2d(num_feat, num_feat // compress_ratio, 3, 1, 1),nn.GELU(),nn.Conv2d(num_feat // compress_ratio, num_feat, 3, 1, 1),ChannelAttention(num_feat, squeeze_factor))def forward(self, x):return self.cab(x)class Mlp(nn.Module):def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.):super().__init__()out_features = out_features or in_featureshidden_features = hidden_features or in_featuresself.fc1 = nn.Linear(in_features, hidden_features)self.act = act_layer()self.fc2 = nn.Linear(hidden_features, out_features)self.drop = nn.Dropout(drop)def forward(self, x):x = self.fc1(x)x = self.act(x)x = self.drop(x)x = self.fc2(x)x = self.drop(x)return xclass DynamicPosBias(nn.Module):def __init__(self, dim, num_heads):super().__init__()self.num_heads = num_headsself.pos_dim = dim // 4self.pos_proj = nn.Linear(2, self.pos_dim)self.pos1 = nn.Sequential(nn.LayerNorm(self.pos_dim),nn.ReLU(inplace=True),nn.Linear(self.pos_dim, self.pos_dim),)self.pos2 = nn.Sequential(nn.LayerNorm(self.pos_dim),nn.ReLU(inplace=True),nn.Linear(self.pos_dim, self.pos_dim))self.pos3 = nn.Sequential(nn.LayerNorm(self.pos_dim),nn.ReLU(inplace=True),nn.Linear(self.pos_dim, self.num_heads))def forward(self, biases):pos = self.pos3(self.pos2(self.pos1(self.pos_proj(biases))))return posdef flops(self, N):flops = N * 2 * self.pos_dimflops += N * self.pos_dim * self.pos_dimflops += N * self.pos_dim * self.pos_dimflops += N * self.pos_dim * self.num_headsreturn flopsclass Attention(nn.Module):r""" Multi-head self attention module with dynamic position bias.Args:dim (int): Number of input channels.num_heads (int): Number of attention heads.qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: Trueqk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if setattn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0proj_drop (float, optional): Dropout ratio of output. Default: 0.0"""def __init__(self, dim, num_heads, qkv_bias=True, qk_scale=None, attn_drop=0., proj_drop=0.,position_bias=True):super().__init__()self.dim = dimself.num_heads = num_headshead_dim = dim // num_headsself.scale = qk_scale or head_dim ** -0.5self.position_bias = position_biasif self.position_bias:self.pos = DynamicPosBias(self.dim // 4, self.num_heads)self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)self.attn_drop = nn.Dropout(attn_drop)self.proj = nn.Linear(dim, dim)self.proj_drop = nn.Dropout(proj_drop)self.softmax = nn.Softmax(dim=-1)def forward(self, x, H, W, mask=None):"""Args:x: input features with shape of (num_groups*B, N, C)mask: (0/-inf) mask with shape of (num_groups, Gh*Gw, Gh*Gw) or NoneH: height of each groupW: width of each group"""group_size = (H, W)B_, N, C = x.shapeassert H * W == Nqkv = self.qkv(x).reshape(B_, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4).contiguous()q, k, v = qkv[0], qkv[1], qkv[2]q = q * self.scaleattn = (q @ k.transpose(-2, -1)) # (B_, self.num_heads, N, N), N = H*Wif self.position_bias:# generate mother-setposition_bias_h = torch.arange(1 - group_size[0], group_size[0], device=attn.device)position_bias_w = torch.arange(1 - group_size[1], group_size[1], device=attn.device)biases = torch.stack(torch.meshgrid([position_bias_h, position_bias_w])) # 2, 2Gh-1, 2W2-1biases = biases.flatten(1).transpose(0, 1).contiguous().float() # (2h-1)*(2w-1) 2# get pair-wise relative position index for each token inside the windowcoords_h = torch.arange(group_size[0], device=attn.device)coords_w = torch.arange(group_size[1], device=attn.device)coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Gh, Gwcoords_flatten = torch.flatten(coords, 1) # 2, Gh*Gwrelative_coords = coords_flatten[:, :, None] - coords_flatten[:, None, :] # 2, Gh*Gw, Gh*Gwrelative_coords = relative_coords.permute(1, 2, 0).contiguous() # Gh*Gw, Gh*Gw, 2relative_coords[:, :, 0] += group_size[0] - 1 # shift to start from 0relative_coords[:, :, 1] += group_size[1] - 1relative_coords[:, :, 0] *= 2 * group_size[1] - 1relative_position_index = relative_coords.sum(-1) # Gh*Gw, Gh*Gwpos = self.pos(biases) # 2Gh-1 * 2Gw-1, heads# select position biasrelative_position_bias = pos[relative_position_index.view(-1)].view(group_size[0] * group_size[1], group_size[0] * group_size[1], -1) # Gh*Gw,Gh*Gw,nHrelative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous() # nH, Gh*Gw, Gh*Gwattn = attn + relative_position_bias.unsqueeze(0)if mask is not None:nP = mask.shape[0]attn = attn.view(B_ // nP, nP, self.num_heads, N, N) + mask.unsqueeze(1).unsqueeze(0) # (B, nP, nHead, N, N)attn = attn.view(-1, self.num_heads, N, N)attn = self.softmax(attn)else:attn = self.softmax(attn)attn = self.attn_drop(attn)x = (attn @ v).transpose(1, 2).reshape(B_, N, C)x = self.proj(x)x = self.proj_drop(x)return xclass SS2D(nn.Module):def __init__(self,d_model,d_state=16,d_conv=3,expand=2.,dt_rank="auto",dt_min=0.001,dt_max=0.1,dt_init="random",dt_scale=1.0,dt_init_floor=1e-4,dropout=0.,conv_bias=True,bias=False,device=None,dtype=None,**kwargs,):factory_kwargs = {"device": device, "dtype": dtype}super().__init__()self.d_model = d_modelself.d_state = d_stateself.d_conv = d_convself.expand = expandself.d_inner = int(self.expand * self.d_model)self.dt_rank = math.ceil(self.d_model / 16) if dt_rank == "auto" else dt_rankself.in_proj = nn.Linear(self.d_model, self.d_inner * 2, bias=bias, **factory_kwargs)self.conv2d = nn.Conv2d(in_channels=self.d_inner,out_channels=self.d_inner,groups=self.d_inner,bias=conv_bias,kernel_size=d_conv,padding=(d_conv - 1) // 2,**factory_kwargs,)self.act = nn.SiLU()self.x_proj = (nn.Linear(self.d_inner, (self.dt_rank + self.d_state * 2), bias=False, **factory_kwargs),nn.Linear(self.d_inner, (self.dt_rank + self.d_state * 2), bias=False, **factory_kwargs),nn.Linear(self.d_inner, (self.dt_rank + self.d_state * 2), bias=False, **factory_kwargs),nn.Linear(self.d_inner, (self.dt_rank + self.d_state * 2), bias=False, **factory_kwargs),)self.x_proj_weight = nn.Parameter(torch.stack([t.weight for t in self.x_proj], dim=0)) # (K=4, N, inner)del self.x_projself.dt_projs = (self.dt_init(self.dt_rank, self.d_inner, dt_scale, dt_init, dt_min, dt_max, dt_init_floor,**factory_kwargs),self.dt_init(self.dt_rank, self.d_inner, dt_scale, dt_init, dt_min, dt_max, dt_init_floor,**factory_kwargs),self.dt_init(self.dt_rank, self.d_inner, dt_scale, dt_init, dt_min, dt_max, dt_init_floor,**factory_kwargs),self.dt_init(self.dt_rank, self.d_inner, dt_scale, dt_init, dt_min, dt_max, dt_init_floor,**factory_kwargs),)self.dt_projs_weight = nn.Parameter(torch.stack([t.weight for t in self.dt_projs], dim=0)) # (K=4, inner, rank)self.dt_projs_bias = nn.Parameter(torch.stack([t.bias for t in self.dt_projs], dim=0)) # (K=4, inner)del self.dt_projsself.A_logs = self.A_log_init(self.d_state, self.d_inner, copies=4, merge=True) # (K=4, D, N)self.Ds = self.D_init(self.d_inner, copies=4, merge=True) # (K=4, D, N)self.selective_scan = selective_scan_fnself.out_norm = nn.LayerNorm(self.d_inner)self.out_proj = nn.Linear(self.d_inner, self.d_model, bias=bias, **factory_kwargs)self.dropout = nn.Dropout(dropout) if dropout > 0. else None@staticmethoddef dt_init(dt_rank, d_inner, dt_scale=1.0, dt_init="random", dt_min=0.001, dt_max=0.1, dt_init_floor=1e-4,**factory_kwargs):dt_proj = nn.Linear(dt_rank, d_inner, bias=True, **factory_kwargs)# Initialize special dt projection to preserve variance at initializationdt_init_std = dt_rank ** -0.5 * dt_scaleif dt_init == "constant":nn.init.constant_(dt_proj.weight, dt_init_std)elif dt_init == "random":nn.init.uniform_(dt_proj.weight, -dt_init_std, dt_init_std)else:raise NotImplementedError# Initialize dt bias so that F.softplus(dt_bias) is between dt_min and dt_maxdt = torch.exp(torch.rand(d_inner, **factory_kwargs) * (math.log(dt_max) - math.log(dt_min))+ math.log(dt_min)).clamp(min=dt_init_floor)# Inverse of softplus: https://github.com/pytorch/pytorch/issues/72759inv_dt = dt + torch.log(-torch.expm1(-dt))with torch.no_grad():dt_proj.bias.copy_(inv_dt)# Our initialization would set all Linear.bias to zero, need to mark this one as _no_reinitdt_proj.bias._no_reinit = Truereturn dt_proj@staticmethoddef A_log_init(d_state, d_inner, copies=1, device=None, merge=True):# S4D real initializationA = repeat(torch.arange(1, d_state + 1, dtype=torch.float32, device=device),"n -> d n",d=d_inner,).contiguous()A_log = torch.log(A) # Keep A_log in fp32if copies > 1:A_log = repeat(A_log, "d n -> r d n", r=copies)if merge:A_log = A_log.flatten(0, 1)A_log = nn.Parameter(A_log)A_log._no_weight_decay = Truereturn A_log@staticmethoddef D_init(d_inner, copies=1, device=None, merge=True):# D "skip" parameterD = torch.ones(d_inner, device=device)if copies > 1:D = repeat(D, "n1 -> r n1", r=copies)if merge:D = D.flatten(0, 1)D = nn.Parameter(D) # Keep in fp32D._no_weight_decay = Truereturn Ddef forward_core(self, x: torch.Tensor):B, C, H, W = x.shapeL = H * WK = 4x_hwwh = torch.stack([x.view(B, -1, L), torch.transpose(x, dim0=2, dim1=3).contiguous().view(B, -1, L)], dim=1).view(B, 2, -1, L)xs = torch.cat([x_hwwh, torch.flip(x_hwwh, dims=[-1])], dim=1) # (1, 4, 192, 3136)x_dbl = torch.einsum("b k d l, k c d -> b k c l", xs.view(B, K, -1, L), self.x_proj_weight)dts, Bs, Cs = torch.split(x_dbl, [self.dt_rank, self.d_state, self.d_state], dim=2)dts = torch.einsum("b k r l, k d r -> b k d l", dts.view(B, K, -1, L), self.dt_projs_weight)xs = xs.float().view(B, -1, L)dts = dts.contiguous().float().view(B, -1, L) # (b, k * d, l)Bs = Bs.float().view(B, K, -1, L)Cs = Cs.float().view(B, K, -1, L) # (b, k, d_state, l)Ds = self.Ds.float().view(-1)As = -torch.exp(self.A_logs.float()).view(-1, self.d_state)dt_projs_bias = self.dt_projs_bias.float().view(-1) # (k * d)out_y = self.selective_scan(xs, dts,As, Bs, Cs, Ds, z=None,delta_bias=dt_projs_bias,delta_softplus=True,return_last_state=False,).view(B, K, -1, L)assert out_y.dtype == torch.floatinv_y = torch.flip(out_y[:, 2:4], dims=[-1]).view(B, 2, -1, L)wh_y = torch.transpose(out_y[:, 1].view(B, -1, W, H), dim0=2, dim1=3).contiguous().view(B, -1, L)invwh_y = torch.transpose(inv_y[:, 1].view(B, -1, W, H), dim0=2, dim1=3).contiguous().view(B, -1, L)return out_y[:, 0], inv_y[:, 0], wh_y, invwh_ydef forward(self, x: torch.Tensor, **kwargs):B, H, W, C = x.shapexz = self.in_proj(x)x, z = xz.chunk(2, dim=-1)x = x.permute(0, 3, 1, 2).contiguous()x = self.act(self.conv2d(x))y1, y2, y3, y4 = self.forward_core(x)assert y1.dtype == torch.float32y = y1 + y2 + y3 + y4y = torch.transpose(y, dim0=1, dim1=2).contiguous().view(B, H, W, -1)y = self.out_norm(y)y = y * F.silu(z)out = self.out_proj(y)if self.dropout is not None:out = self.dropout(out)return outclass VSSBlock(nn.Module):def __init__(self,hidden_dim: int = 0,drop_path: float = 0,norm_layer: Callable[..., torch.nn.Module] = partial(nn.LayerNorm, eps=1e-6),attn_drop_rate: float = 0,d_state: int = 16,expand: float = 2.,is_light_sr: bool = False,**kwargs,):super().__init__()self.ln_1 = norm_layer(hidden_dim)self.self_attention = SS2D(d_model=hidden_dim, d_state=d_state,expand=expand,dropout=attn_drop_rate, **kwargs)self.drop_path = DropPath(drop_path)self.skip_scale= nn.Parameter(torch.ones(hidden_dim))self.conv_blk = CAB(hidden_dim,is_light_sr)self.ln_2 = nn.LayerNorm(hidden_dim)self.skip_scale2 = nn.Parameter(torch.ones(hidden_dim))def forward(self, input, x_size):# x [B,HW,C]B, L, C = input.shapeinput = input.view(B, *x_size, C).contiguous() # [B,H,W,C]x = self.ln_1(input)x = input*self.skip_scale + self.drop_path(self.self_attention(x))x = x*self.skip_scale2 + self.conv_blk(self.ln_2(x).permute(0, 3, 1, 2).contiguous()).permute(0, 2, 3, 1).contiguous()x = x.view(B, -1, C).contiguous()return xif __name__ == '__main__':# 初始化VSSBlock模块，hidden_dim为128block = VSSBlock(hidden_dim=128, drop_path=0.1, attn_drop_rate=0.1, d_state=16, expand=2.0, is_light_sr=False)# 将模块转移到合适的设备上device = torch.device("cuda" if torch.cuda.is_available() else "cpu")block = block.to(device)# 生成随机输入张量，尺寸为[B, H*W, C]，这里模拟的是批次大小为4，每个图像的尺寸是32x32，通道数为128B, H, W, C = 4, 32, 32, 128input_tensor = torch.rand(B, H * W, C).to(device)# 计算输出output_tensor = block(input_tensor, (H, W))# 打印输入和输出张量的尺寸print("Input tensor size:", input_tensor.size())print("Output tensor size:", output_tensor.size())
运行结果
第三章：经典文献阅读与追踪
Mamba原文：Mamba: Linear-Time Sequence Modeling with Selective State Spaces
经典论文
- Vision Mamba@Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model
- MambaIR@MambaIR: A Simple Baseline for Image Restoration with State-Space Model
- U-Mamba@U-Mamba: Enhancing Long-range Dependency for Biomedical Image Segmentation
Mamba_1154">Mamba系列论文追踪
该Github链接会分享不同领域基于Mamba结构的论文
Mamba_State_Space_Model_Paper_List Public：https://github.com/Event-AHU/Mamba_State_Space_Model_Paper_List
Mamba_1167">第四章：Mamba理论与分析
FusionMamba: Efficient Image Fusion with State Space Model【文献阅读】
Mamba_1173">Mamba模块
借用该论文的图3来一起学习一下Mamba模块的结构：
其中，最左边的就是Mamba模块。Vision Mamba模块要对特征图进行特征提取。因此，我们期望经过Mamba模块后的特征图的大小不变。
第一部分：把输入的特征图F_in，其维度为H，W，C送入LayerNorm层，映射得到两个不同的特征X和Z，它们的维度不变为H，W，C。
第二部分：对X沿着4个不同的方向进行Fatten展平得到1维的特征向量，这4个方向特征向量的维度是HW，C这儿和Transformer的变换类似，转换成TOKEN，然后再去进行后续计算。4个不同方向的展平方式，如上图最右边所示，就是从左到右、从上到下四个方向。
第三部分：将4个不同方向的1维特征向量送入SSM模块进行特征提取，看来SSM模块就是Mamba模块的核心了，这个我们将在后文对它进行详细的解读。
第四部分：将输出的特征向量其维度为HW，C，经过unflatten就是还原成特征图维度为H，W，C后将4个方向的特征图加起来，进行充分的融合得到特征Y。
第五部分：对最初的特征Z经过SiLU进行非线性映射，作为权重或者注意力与融合的特征图Y进行激活或者加权得到显著性的特征。最后将特征经过1×1的卷积进行映射后与输入的特征做一个残差得到最终的输出特征F_out。
关键的SSM算法
按照该论文给出的流程图，我们来对SSM算法进行一个充分的理解。如下图最左边，右边不用管是作者对其的改进。
SSM Block未完待续...
第五章：总结和展望
- 2024年04月29日16:57:45，今天已完成环境的安装与即插即用模块实例化和相关论文的分享；在近期会充分学习Mamba后对其理论进行分享，帮助快速简要理解原文Mamba相关理论。
- 2024年05月02日15:56:32，今天基于一篇FusionMamba的论文补充了Mamba模块的基础知识，后面将重点介绍其中的SSM模块，就会完成本博客的分享。
